vhdl code for exponent value

[ganadikamal]

hi i want to write a code for 1/(1+e^(-x)).. i implemented using the value of e like 2.134 the problem is in the power i want to use the floating point .. how can i solve this.. please help in sort outing this problem..

 

[FvM]

In most cases, you would prefer fixed point number formats to true floating point. But it's a matter of the application requirement. You need to become clear about number ranges and required resolution.

For the exponential function, you can consider two options:
- a ROM table, which could also represent the complete function
- classical iterative methods known from numerical mathematics text books, e.g. geometric series

 

[ganadikamal]

thank u for replying. let me try that..